The use of Hyperspectral Imaging (HSI) for Digital Lithological Mapping is becoming a crucial turning point in the area of Geological Science.  Through this research, we try to develop algorithms capable of identifying probable surface mineral deposits by the use of HSI incorporating Machine Learning and Signal Processing approaches. Currently, I am developing an algorithm capable of identifying survey sites for Ilmenite, Limestone, and Montmorillonite deposits through site-specific endmember extraction in different regions of Sri Lanka. 

 This research is carried out under the "International Development Research Center, Canada" grant as a joint research of Sri Lanka and Malaysia to build an Artificial Intelligence framework for threat assessment and containment for COVID-19 and future epidemics while mitigating the socioeconomic impact on women and children, and underprivileged groups. An island-wide survey was carried out to assess the effectiveness and impact of different containment strategies. Furthermore, the team was able to build a simulator to analyze human mobility patterns, disease propagation patterns, and asses the effectiveness of different containment strategies. Currently, I am working on modeling the simulator to mimic the real-world scenario by tuning it through the data collected from the survey.  In addition, I am working on identifying patterns in human mobility with respect to their occupations through GPS trajectories. 

The contamination on High Voltage (HV) insulator surfaces in transmission and distribution power lines increases the possibility of flashover. The common contaminants on HV insulators can be identified as Dust, Algae, Salts, and Industrial Contaminants. This research focuses on finding algae and salt contaminations through the use of Machine Learning and Signal Processing by utilizing RGB, Thermal, and Multispectral Images. An algorithm was developed to identify the algae-affected regions of HV insulators using the Bhattacharyya Distance. Currently, the research focuses on identifying different salt contamination concentrations under different humidity conditions to asses the insulation condition. 

Fish quality assessment is a vital process in the fisheries industry. It is necessary to identify the quality of the fish in terms of the degree of degradation prior to it being sent to the consumer. In this research, we focus on assessing the quality of yellowfin tuna, Mackerel Tuna, and Spotted Sardinella using Multispectral and RGB images. For this several image processing, deep learning, and machine, learning-based approaches are being exploited. 

A frequently used edible oil is coconut oil, which is derived from the coconut kernel. This has been one of the primary ingredients in the food industry due to its numerous health advantages and nutritional value. However, adulterating coconut oil using different edible oils has reduced its quality causing major health issues. Through this research, a principal component analysis based Bhattacharyya distance model is developed to identify the adulteration level of coconut oil where reused coconut oil is being used as the adulterant.